research open access hypocrea jecorina cel6a protein ... · research open access hypocrea jecorina...

RESEARCH Open Access

Hypocrea jecorina CEL6A protein engineeringSuzanne E Lantz1*, Frits Goedegebuur2, Ronald Hommes2, Thijs Kaper1, Bradley R Kelemen1, Colin Mitchinson1,Louise Wallace1, Jerry Ståhlberg3, Edmundo A Larenas1

Abstract

The complex technology of converting lignocellulose to fuels such as ethanol has advanced rapidly over the pastfew years, and enzymes are a critical component of this technology. The production of effective enzyme systemsat cost structures that facilitate commercial processes has been the focus of research for many years. Towards thisend, the H. jecorina cellobiohydrolases, CEL7A and CEL6A, have been the subject of protein engineering at Genen-cor. Our first rounds of cellobiohydrolase engineering were directed towards improving the thermostability of bothof these enzymes and produced variants of CEL7A and CEL6A with apparent melting temperatures above 70°C,placing their stability on par with that of H. jecorina CEL5A (EG2) and CEL3A (BGL1). We have now moved towardsimproving CEL6A- and CEL7A-specific performance in the context of a complete enzyme system under industriallyrelevant conditions. Achievement of these goals required development of new screening strategies and tools. Wediscuss these advances along with some results, focusing mainly on engineering of CEL6A.

BackgroundHypocrea jecorina (anamorph Trichoderma reesei) is anindustrially important producer of cellulases for anapplications portfolio that includes pulp and paper pro-cessing, food and feed processing, textile manufacturingand modification, and detergents/cleaners formulation.Cellulases were employed in early ‘waste to transporta-tion fuel’ research in, for example, the U.S. Army Natickprogram in the 1970s; however, it was not until 2007that a cellulase formulation for biomass saccharificationwas developed and marketed to the nascent biomass-to-ethanol industry (Accellerase®1000; Genencor Division,Danisco USA Inc.).

MechanismTo understand the challenges of protein engineeringbiomass enzymes, it is important to understand currentconcepts of their structure-function relationship. Earlyconcepts of cellulase mode of action were published byReese et al. in 1950 [1], followed by a review article in1964 [2]. Since then, a great deal of work has beendevoted to elucidating the mechanism of enzymatic cel-lulose degradation. These studies have covered a widerange of substrates from pure cellulose to pretreated

lignocellulose from various sources [3-8]. Yet, there isstill much to learn regarding the fundamental mechan-ism of cellulases.Two of the most important cellulase glycosyl hydrolase

families are GH6 and GH7. The first structure of a cello-biohydrolase to be solved was the catalytic module ofH. jecorina CEL6A (previously known as CBH 2), whichwas published by Rouvinen et al. in 1990 [9]. The firststructure of a fungal cellulose binding module, the CBM1of H. jecorina CEL7A (previously known as CBH 1), waspublished by Kraulis et al. in 1989 [10], and a few yearslater the structure of the H. jecorina CEL7A catalyticmodule was published by Divne et al. in 1994 [11].The catalytic mechanism of glycosyl hydrolase family 6

enzymes, including both exo- and endoglucanases is netinversion (versus retention in family 7) of the anomericconfiguration(Carbohydrate Active Enzymes databasehttp://www.cazy.org/; Cantarel [12]). Although there is nooverall sequence homology between the catalytic domainsof the H. jecorina cellobiohydrolases CEL6A and CEL7A,there is homology between CEL6A and GH6 endogluca-nases and between CEL7A and GH7 endoglucanases. It isthe differences, particularly in the loop regions that formthe active site tunnel in CEL6A and CEL7A, that affect thetype of enzyme activity. These loops are missing in theendoglucanases, leaving a more open active site cleft[9,11]. Indeed, deletion of the C-proximal loop that coversthe active site of Cellulomonas fimi cellobiohydrolase

* Correspondence: [email protected] Division, Danisco USA Inc., 925 Page Mill Rd. Palo Alto, CA 94304,USAFull list of author information is available at the end of the article

Lantz et al. Biotechnology for Biofuels 2010, 3:20http://www.biotechnologyforbiofuels.com/content/3/1/20

© 2010 Lantz et al; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative CommonsAttribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction inany medium, provided the original work is properly cited.

http://www.cazy.org/

mailto:[email protected]

http://creativecommons.org/licenses/by/2.0

A increased its endoglucanase activity [13]. Furthermore,Divne et al. [11] concluded that dissimilarities in theCEL6A and CEL7A tunnels (length, number of trypto-phans, and asymmetric distribution of glucose bindingsubsites) could explain observations of distribution differ-ences on cellulose fibrils and synergy between CEL7A andCEL6A.Advanced modeling [14,15] and microscopy techni-

ques [16,17] are being used to study cellobiohydrolase-substrate interactions. These techniques hold promisefor demonstrating how these enzymes bind substrateand move and how these events are related to hydroly-sis. Studies of the H. insolens CEL6A [18] provided adetailed description of solvent-mediated carbohydrate-protein interactions involved in ligand movementthrough the active site tunnel during hydrolysis usingcrystal structures to support the authors’ model. Elec-tron microscopy studies with enzymes from Humicolainsolens confirmed that the GH6 and GH7 cellobiohy-drolases act processively from opposite ends of the cel-lulose chain, GH6 from the nonreducing end and GH7from the reducing end [19]. These studies also revealedthat CEL6A had substantial endoglucanase activity andlower processivity than CEL7A, at least on the Valoniacellulose crystals studied. However, on a complex bio-mass substrate, in a natural mix of cellulases, it is notknown to what extent CEL6A relies on its own endoglu-canase activity and to what extent it starts from chainends created by other endoglucanase enzymes. TheWalker lab at Cornell University uses single-moleculedetection methods, such as quantitative fluorescencemicroscopy, to study cellulase adsorption and hydrolysis[20]. Recently, researchers have observed cellulases slid-ing along a cellulose fibril in real time and measured thevelocity of sliding [21]. While the basic chemicalmechanisms of glycosyl hydrolases are well understoodand their combined roles in the hydrolysis of pure crys-talline cellulose are reasonably well elucidated, hydroly-sis of lignocellulosic substrates is more complicated andis the subject of continuing research. Much of theapplied research is now directed towards discovering theminimum number of enzyme components that provideoptimal performance.

Optimal cellulase mixture for biomass hydrolysisTo date, optimal mixtures have generally been devel-oped by purely empirical approaches [22,23] and mayvary relative to the substrate being used. The CEL7Aand CEL6A cellobiohydrolases are the two most abun-dant enzymes in H. jecorina [23], indicating their keyrole in the cellulase enzyme system. The importance ofCEL7A and CEL6A was demonstrated in 1980 by Reeseand Mandels [24] when they showed that cellobiohydro-lase activity was limiting cellulose hydrolysis. They also

demonstrated that the cellobiohydrolase component wasless stable under standard process conditions (pH 4.8,50°C, 24 hr) than most of the other enzymes in thenative H. jecorina cellulase milieu. That the cellobiohy-drolases are indeed key players was further supported byindividual deletion of the four major cellulase genescbh1, cbh2, egl1 and egl2 (coding for CEL7A, CEL6A,CEL7B and CEL5A) in the industrial H. jecorina strainVTT-D-79125, where the mutants without CEL7A orCEL6A showed a 70% and 33% activity loss, respectively,on filter paper relative to the parent strain [25]. Thisstudy, and that of Reese and Mandels [24], was con-ducted with pure cellulose substrates (filter paper andAvicel, respectively). Rosgaard et al. [23], using two dif-ferent hot water-pretreated barley samples, demon-strated optimal performance when the CEL6A:CEL7Aratio was 2:1, again showing that cellobiohydrolase activ-ity exceeding that of native H. jecorina preparations wasbeneficial.The issues of inhibition (substrate and product) and

substrate accessibility in these complex mixtures havebeen recognized for some time [26-29]. It was alsoaccepted early on that efficient cellulose-digestingenzymes would need to be produced with acceptableeconomics. Much effort has been made to improve theproductivity of strains descending from the original T.reesei QM6a [30]. Programs such as the NTG mutagen-esis and selection conducted at Rutgers University [31]and Lehigh University [32] were quite successful andresulted in strains that are still in industrial use, such asRutC30 and RL-P37, and their descendants. Strainimprovement has continued even with the relativelyhigh productivity of these industrial strains. To date,enzyme cost reduction has been mainly accomplished bystrain improvement rather than by protein engineering.

Protein engineeringThe generally low turnover rates of cellulases, for exam-ple, one to four per second for Cel7A [33], present achallenge in applications where cheap sugars are thedesired product. Faster rates may be accomplished byraising the reaction temperature and/or by increasingthe specific activity of the enzyme. CEL6A has been thetarget of several protein engineering efforts; some weredescribed in Schülein’s 2000 review [34] of cellulaseengineering, where he acknowledged that most of thework was devoted to understanding the catalyticmechanism. At VTT, Koivula et al. [35] identified theCEL6A catalytic domain surface residue W272 as essen-tial for degradation of crystalline cellulose, but not solu-ble or amorphous cellulose. Wohlfahrt et al. [36]identified carboxyl-carboxylate pair mutations (to thecorresponding amide-carboxylate pair) to be useful instabilizing H. jecorina CEL6A, particularly with respect


of 13

to pH. More site-directed mutations pointed to a rolefor Y169 in kinetics and binding [37]. The Y169F muta-tion resulted in little change in the crystal structure rela-tive to wild-type Cel6A, but the association constantsfor cellotriose and cellotetraose increased fourfold whilethe activity decreased to about one fourth its originallevel. The authors speculate that the Y169 residueimposes a distortion of glucose to a more reactive con-formation in the active site tunnel. Mutations of twocarboxylic acid residues, D175 and D221, at the catalyticcentre of CEL6A, supported the proposed role of D221as responsible for protonation of the glycosidic oxygen,while molecular dynamics simulations indicated thatD175 indirectly fulfills the role of a catalytic base [38].However, in most of the structures of GH6 enzymes,D221 (or the equivalent residue) makes a hydrogenbond to and shares a proton with D175; therefore, D221cannot protonate the glycosidic oxygen. Formation of acatalytically competent configuration seems to be asso-ciated with closing of a flexible loop in association withthe substrate, which isolates two water molecules at thecatalytic centre [39]. D175 is in contact with and mayaccept a proton from one of the water molecules, whichmay in turn accept a proton from the second water thatis in position for nucleophilic attack on the anomericcarbon of the sugar. But the closing of the loop alsorestricts the passage between successive glucose bindingsubsites within the active site, suggesting that processiveaction along a cellulose chain requires a cyclic closing-opening sequence for each hydrolytic event, in confor-mity with molecular dynamics [38]. Flexibility in thisloop may thus be an important factor for the specificactivity of the enzyme.Wilson and colleagues [40-42] have worked exten-

sively with Thermobifida fusca CEL6B exocellulase andhave characterized the catalytic mechanism usingenzyme variants. With CEL6B variants expressed in S.lividans or E. coli, they demonstrated that lower activitywith insoluble substrates was linked to reduced proces-sivity and that adding disulfide bonds across the loopsforming the active site tunnel reduced ligand binding,processivity and activity. In addition, they identifiednoncatalytic CEL6B mutations in which single and dou-ble mutants (G234 S, G284P) demonstrated higheractivity on swollen cellulose and filter paper, but theseimproved variants did not increase synergism with theT. fusca endoglucanase Cel5A [40]. Mutations near thesubstrate binding site were found to reduce cellobioseinhibition, but in most cases (except G234S) the muta-tions also resulted in reduced thermal stability. Theeffect of cysteine residue mutations on expression andthermostability was related to the position of the residue(s) and whether it led to aberrant disulfide bond forma-tion, improper folding, and sometimes proteolysis [41].

Wilson described these and other complexities of engi-neering cellulases for enhanced activity in a 2009 reviewarticle [42]. The review cautions that (1) there is adearth of demonstrated cellulase activity improvement,(2) improving a single activity may be irrelevant if per-formance of the synergistic mixture is not improved, (3)protein engineering using only the catalytic module doesnot guarantee the same performance in the full lengthprotein and (4) activity improvement demonstrated onone substrate does not guarantee the same results on adifferent substrate.Recently, Heinzelman and colleagues [43] demon-

strated that CEL6A chimeras, which included sequencesfrom H. jecorina CEL6A, CEL6A from the thermophilicfungus Humicola insolens, and/or sequences from threeother fungi, expressed by S. cerevisiae, resulted in mole-cules with greater thermostability than either parent.The chimeras demonstrated a broader pH range than H.jecorina CEL6A. Of the chimeras described, noneexceeded CEL6A specific activity on phosphoric acidswollen cellulose (PASC). In a related study [44], theyidentified a single mutation that significantly enhancedH. jecorina CEL6A thermal stability with PASC hydroly-sis activity similar to wild type.Yet, after all these years, cellulosic ethanol is still

expensive relative to starch ethanol. The saccharifyingenzymes compose a significant cost component, so it isour objective to reduce the dose required to convertpretreated biomass to simple sugars for fermentation.One way to optimize performance is through proteinengineering. Engineered cellulases have been used intextile applications, but few examples exist in the fieldof biomass saccharification. Unlike textile enzyme pro-ducts which often are monocomponent, lignocellulosicbiomass conversion requires a complex mixture ofenzymes. This presents several challenges to a proteinengineering approach to increasing performance. Forexample, it is not efficient to engineer all of the requiredenzymes. However, within the enzyme complex used forlignocellulosic conversion, some activity is always goingto be either limiting or required in abundance. The pre-ferred protein target(s) for improvement is an enzymethat is limiting and has the potential for making a largeimpact on the specific performance of the system. Thisimprovement could be the result of one or a combina-tion of improvements such as increased specific activity,reduced product inhibition, reduced nonproductivebinding, or enhanced stability and longevity under pro-cess conditions. Studies using pure cellulose and tradi-tional methods for detecting sugar release [45-49] haveprovided insight into enzyme mode of action, but purecellulose does not necessarily predict enzyme perfor-mance on pretreated lignocellulose. Such improvementsshould be measured under process-relevant conditions,


of 13

making development of the appropriate screening assaycritical. Screens conducted in the complex biomassmilieu will enable the detection of variants that are lesssusceptible to inhibition and inactivation from compo-nents found in the biomass of interest.

Results and DiscussionMixture studiesWe demonstrated that the performance optima of a cel-lulase mixture also favoured more cellobiohydrolaseactivity using a process-relevant substrate, dilute acid-pretreated sugarcane bagasse, in a computer-designedmixture experiment (Design Expert Dx5). In this experi-ment, purified H. jecorina CEL7A and CEL6A werecombined in various ratios with an RL-P37 backgroundsample from which the cel7A and cel6A genes had beendeleted (delta delta P37). Samples were dosed such thatthe final cellulose concentration of the mixture was 6%(wt/wt). In the three component mixtures, the amountsof CEL7A, CEL6A and delta delta P37 proteins rangedfrom 5 to 85 weight percent of the total mixture pro-tein. Five percent (wt/wt) H. jecorina Cel3A (beta gluco-sidase 1) was included in all of the mixtures to mitigatethe impact of cellobiose inhibition by converting mostof the soluble cellooligosaccharides to glucose, whichwas analyzed by high-pressure liquid chromatography(HPLC). At both 50°C and 60°C, the performance of themixture favoured more CEL7A and CEL6A activity andin approximately equal proportions, whereas in thenative preparation CEL7A is three- to fourfold moreabundant than CEL6A (Figure 1). The cellobiohydro-lases were more limiting at 60°C than at 50°C, possiblydue to their relatively lower thermostability. Enhancedactivity and thermal stability of cellobiohydrolases arethus target properties to enhance for improvement ofbiomass conversion performance.

Cellulase engineering for thermostabilityFrom 2000-2004, Genencor worked with the NationalRenewable Energy Laboratory (NREL) under the aus-pices of the Department of Energy Office of the BiomassProgram and improved the thermal stability of CEL6Aand CEL7A through protein engineering.Differential scanning calorimetry (DSC) was used to

determine the thermal midpoint (Tm) values of thethermal unfolding process for the most abundantH. jecorina cellulases and b-glucosidase (Table 1).Excessive heat capacity curves were measured using anultrasensitive scanning high-throughput microcalori-meter, VP-Cap DSC (MicroCal, Inc., Northampton,MA). H. jecorina enzymes (500 μl of 0.5 mg/ml) werescanned over 30-90°C temperature range in 10 mMsodium acetate buffer, pH 5.0. A 200°C/hr scan ratewas used. The Tm of the DSC curves was used as anindicator of the thermal stability and calculated usingthe Origin Lab 7.0 software. Native CEL6A and CEL7Aare more thermolabile than CEL5A and CEL3A, whichmay affect whole cellulase performance in extendedsaccharification reactions at >50°C. With the screeningtools available in 2000, protein engineering was under-taken to improve the thermal stability of the twoH. jecorina cellobiohydrolases.Different mutagenesis and screening approaches were

used for CEL7A and CEL6A Tm improvement. Themelting point of CEL7A was increased through a combi-nation of random and site-directed mutagenesis andscreening (US 2007/0173431). A limited number of siteswith potential involvement in stability were selected onthe basis of structure and a 42-member CEL7Asequence alignment of Hypocrea and Trichodermafamily members. Site saturation mutagenesis wasperformed on these sites. CEL7A variants containingfrom 1 to 19 mutations were expressed in A. niger for

Figure 1 Mixture experiment. Three component design of experiment shows limitation and thermolability of cellobiohydrolase activitycompared to the cellulase background of delta-CEL7A delta-CEL6A RL-P37. Shown are the experimental design points (left), contour plots for 50°C saccharification (middle) and 60°C saccharification (right). Red dots represent actual data points. Duplicate data points are indicated by thelabel (2). Contour labels indicate percent glucan conversion.


of 13

screening, and approximately 100,000 clones wereassayed for improved stability. Stability was determinedby the difference in 4-methylumbelliferryl-lactoside(Sigma Chemicals, M2405) activity before and after aheat challenge. Select A. niger-expressed variants werepurified by hydrophobic interaction chromatography,and thermal stability was determined by circular dichro-ism spectrophotometry (CD) [50-52].Point mutations were then combined to obtain

CEL7A variants with substantially higher thermal stabi-lity (Goedegebuur et al., manuscript in preparation).Eighteen sites were combined to produce one CEL7Avariant, which when expressed in T. reesei had a Tm

increase of 14.8°C (Tm 76.0°C) as determined by DSC.While screening was performed on A. niger-expressedproteins, lead molecules were then expressed in T. ree-sei for validation of performance, expression, and Tm

determination.CEL6A variants were also expressed in A. niger for

screening. A limited number of sites were selected formutagenesis through a consensus approach (US20060205042). A sequence alignment of H. jecorinaCEL6A and eight GH6 family members was used toconstruct a consensus sequence. Single and multipleamino acid mutations were designed and made by sitemutagenesis. Nonconserved positions were examined inthe crystal structure, and mutations were selected thatwere different in CEL6A from the consensus and whichfit the structure without disturbance. Conserved siteswere not changed. More than 5000 clones werescreened on PASC (prepared from Avicel PH101 [53])for remaining activity after heat inactivation for 1 hr at61°C or 65°C at pH 4.85. Combinations of mutationswere made and ultimately resulted in an H. jecorina-expressed CEL6A variant with a Tm increase of 6.9°C(as determined by DSC). This thermostable variant wasshown to have similar activity to the CEL6A wild typein a reconstituted whole cellulase in dilute acid pre-treated corn stover (PCS [54]) hydrolysis. PCS (7% wt/wt cellulose) specific performance was tested at 53°C for20 hr by adding CEL6A variants to a CEL6A-deletedcellulose strain product (US 2006/0205042). These two

protein engineering projects resulted in apparent melt-ing temperatures above 70°C for both CEL7A andCEL6A, placing their stability on par with that ofCEL5A and CEL3A.

Cellulase engineering for performanceWith this foundation of knowledge and experience, andwith new tools in hand, we tackled the challenge ofimproving the specific activity of H. jecorina CEL6A.The products of this research will contribute to the costreduction of enzymes in biomass-to-ethanol processesand one example for the application of the improvedenzymes is the demonstration plant that the DuPontDanisco Cellulosic Ethanol, LLC (DDCE) began operat-ing in early 2010.Since our initial protein engineering work with NREL,

we have developed T. reesei as a screening host for pro-tein engineering and made other technical advance-ments on the basis of the lessons learned during thoseearly years. Because T. reesei is a preferred host forenzyme production, we developed it as the host forscreening to ensure that expression and performance ofthe selected variants would not be lost when expressedin the production host. The requirements for an effec-tive high throughput screening strain include high fre-quency of transformation, reliable gene expression,reproducible growth in microtiter plate (MTP) format,sufficient and reproducible protein production, and lowsecretion of background proteins. The latter was espe-cially important because the candidates for protein engi-neering were overexpressed using a cellulase inductionsystem that had the potential to induce production ofconfounding background activities. The strain basis forscreening was a fourfold deletion variant of T. reesei inwhich the genes for cel7A, cel6A, cel7B, and cel5A hadbeen removed.The biomass saccharification assay was miniaturized

from shake flasks to 96-well MTP. Although miniatur-ized, the assays incorporated process relevant condi-tions. Either washed or unwashed pretreated biomasswas used as the substrate. The substrate was deliveredto the MTP as slurry in pH 5 sodium acetate buffer,with a consistency like cake batter. We demonstratedthat the MTP scale assay was predictive of shake flaskscale results (Figure 2). We also found that the MTPscale assay was predictive of larger-scale performance(not shown). MTP and shake flask saccharificationassays were incubated for 3-5 days at 50°C, with shak-ing, using washed PCS at 13% wt/wt final solids (7% wt/wt cellulose). The correlation was shown by comparisonof MTP scale results with shake flask scale results fromNREL using the same materials and conditions http://www.nrel.gov/biomass/pdfs/42629.pdf. Each scale pre-sented different challenges in delivery, mixing, and

Table 1 H. jecorina cellulases

Enzyme Class Wt/wt% Tm (°C)

CEL7A (CBH I) EXO 40-60 61.2

CEL6A (CBH II) EXO 12-20 67.2

CEL7B (EGL I) ENDO 5-10 67.6

CEL5A (EGL II) ENDO 1-10 72.5

CEL12A (EGL III) ENDO < 1-5 63.0

CEL3A (BGL1) b-glucosidase 1-2 77.0

Relative abundance [23] and Tm values for the thermal unfolding processmeasured by differential scanning calorimetry in 10 mM acetate buffer, pH5.0, of H. jecorina cellulases and b-glucosidase are shown.


of 13

http://www.nrel.gov/biomass/pdfs/42629.pdf

http://www.nrel.gov/biomass/pdfs/42629.pdf

sampling. It is critical that the small-scale assays predictlarge-scale results. Screens performed with pure cellulo-sic substrates are often not predictive and do not allowfor a mechanistic understanding of complex substrates.In fact, the results of screening with a particular com-plex substrate may not accurately predict performanceon a different complex substrate. This is illustrated in

the results of a performance comparison of 62 indepen-dent samples of T. reesei whole cellulase in saccharifica-tion assays with four different substrates: dilute acidpretreated sugarcane bagasse, PCS, Avicel, and PASC.The dilute acid PCS and bagasse were produced andprovided by NREL [54]. The composition of lignocellu-losic materials was determined using the assays detailedin the NREL protocols for Standard Biomass AnalyticalProcedures http://www1.eere.energy.gov/biomass/analy-tical_procedures.html.The 62 cellulase samples represented material from

various Genencor T. reesei strains, production lots, pro-tein production conditions, and formulations, collectedover several years. In the performance assays, the cellu-lases were dosed at 20 mg total protein per gram of cel-lulose. Total protein was determined by an automatedBiuret method (Pointe Scientific T7528). The substrateloading of the PCS, bagasse, and Avicel was 7% (wt/wt)cellulose. Substrates were incubated with the cellulasesfor 3 days at 50°C, pH 5, and 200 rpm shaking. The cel-lulases were incubated with 1% (wt/wt) PASC at 50°C,pH 5, with shaking for 1 hr. Cellulose hydrolysis wasmeasured either by a reducing sugar release assay (e.g.,PAHBAH assay [49]), or by HPLC. Although clean cel-lulose such as Avicel and PASC generally correlatedwith lignocellulose conversion, there were exceptions(Figure 3). For example, sample 35 showed overall goodperformance on all four substrates; however, sample 57performed well on PASC and Avicel and poorly onbagasse and PCS. Overall, enzyme performance wasgreater on bagasse than on PCS. There was little corre-lation between performance and enzyme productionprocess or formulation or sample age. On the basis of

Figure 2 Saccharification assays. The miniaturized saccharificationassay with dilute acid pretreated corn stover is predictive of shakeflask scale performance. The shake flask assay was conducted byNREL according to their Laboratory Analytical Procedure (LAP)“Enzymatic Saccharification of Lignocellulosic Biomass” http://www.nrel.gov/biomass/pdfs/42629.pdf.

Figure 3 H. jecorina cellulase performance. Comparison of saccharification performance (glucan conversion) on biomass and model cellulosicsubstrates.


of 13

http://www1.eere.energy.gov/biomass/analytical_procedures.html

http://www1.eere.energy.gov/biomass/analytical_procedures.html

http://www.biotechnologyforbiofuels.com/content/3/1/20

http://www.biotechnologyforbiofuels.com/content/3/1/20

previous experience, all samples were assumed to bestable during storage.Because of these results, our screening assays for pro-

tein engineering were developed to bring them closer toactual use conditions. This required the development ofbiomass performance screens in which the activity of aspecific enzyme or variant could be queried within a cel-lulase background. The challenge of screening for thetarget cellulase activity in a background of other cellu-lases is not trivial due to the synergistic nature of theenzymes. Another advance was using pretreated ligno-cellulosic substrate at high solids. Screening for CEL6Aspecific activity improvements required development oftwo high-throughput assays: one that showed dosedependence with respect to CEL6A concentration andone to accurately determine the concentration ofexpressed CEL6A. Variants were screened in a reconsti-tuted cellulase background lacking cellobiohydrolaseactivity and including sufficient b-glucosidase activity toproduce primarily glucose, which was measured byreducing sugar analysis using the PAHBAH method[49]. The substrate was washed PCS. Specific activityscreening became possible with HPLC determination ofprotein concentrations in a 96-well MTP format.Although specific activity was the target property forimprovement, stability using PASC was also measured.Our protein engineering approach was to create Site

Evaluation Libraries (SELs) that contained all 19 aminoacid substitutions (including recreation of wild type).The libraries were generated in E. coli, variants weresequenced, and plasmid DNA was transformed into T.reesei for expression and screening. Each variant wasscreened for multiple properties to ensure that impor-tant properties, such as thermal stability and perfor-mance, were not lost.Selection of the CEL6A sites for engineering was based

on knowledge of the enzyme structure and guided bysequence alignments. More than 100 nonconservedCEL6A residues were selected for mutagenesis. They cov-ered about 30% of the molecule concentrating on cataly-tic domain surface residues, but also including sites inthe linker region and the carbohydrate binding module.Active site residues were not targeted in this study.Although it was tempting to use a tagged molecule for

ease of separation of the protein of interest from thecellulase background, we demonstrated that a C-term-inal His tag on CEL6A caused reduced performance inbiomass assays. In contrast, PASC assay performance ofCEL6A was not affected by the tag, which emphasizedagain the need to use process-relevant conditions inscreens (Figure 4). Instead, we developed proprietaryhigh-throughput assays for HPLC determination ofCEL6A and variant protein concentration to enable

calculation of specific activities and dose-dependentbiomass performance of CEL6A.MTP-scale saccharification assays using PCS and

PASC were used to screen the CEL6A variants. Serialdilutions of the CEL6A variants were added to the PCSsaccharification assay such that a dose-response curvecould be generated. Cellulose hydrolysis was determinedby measurement of the increase in reducing sugars(PAHBAH). A performance index (PI) was calculatedfor each variant. The performance index is the ratio ofperformance of the variant to the wild-type protein.Generally an improved variant would have a PI >1, asshown in Figure 5. Although saccharification data is notlinear with respect to CEL6A concentration and requiresa curve fit, improved variants can be detected with thisassay and analysis. CEL6A controls were included ineach MTP in two formats. Plate-to-plate reproducibilityof wild-type CEL6A was compared for PASC perfor-mance and found to be acceptable for detecting winners.Each growth plate contained recreated wild types in theSELs as well as wild-type controls. In Figure 6, eachgraph shows the activity of the wild-types on eachPASC assay plate plotted against the isotherm fit for theactivity of the wild-types from all 34 plates (blue line).The same comparison was made for PCS and found tobe acceptable. In addition to the PCS and PASC sac-charification assays, CEL6A variants were also screenedfor ethanol stability and heat stability (PASC activitybefore and after heating at a challenge temperature).For each assay, we graphed the natural log of the PI

for wild-type performance (those that were recreatedwithin each site library). The transformed data are aGaussian distribution (Figure 7) centered on zero foreach property. These wild-type transformants were notused to calculate the PI. Two types of wild-type trans-formants were on each library plate: those that wereused to calculate the PI (the controls) and the recreatedwild types that were used to test the curve fit. TheGaussian distribution of the wild types was as expected.A correlation was observed between the two activity

assays, PCS and PASC, from the graph of the naturallog of the PI for all of the transformants (wild type andvariants) (Figure 8). The variants in the upper rightquadrant were improved approximately 2.5-fold overwild type. One false-positive wild type was observed inthis quadrant. A correlation was also observed betweenthe two stability assays: heat and ethanol (data notshown). There was little correlation between perfor-mance in the PASC activity assay and either stabilityassay (Figure 9).The performance data can be sorted in a variety of

ways, depending on the query objective. The most strin-gent case would be selection of mutations that resultedin improvements over wild type in all four assays


of 13

(ethanol stability, heat stability, PASC activity and PCSactivity). A less stringent selection would be for wild-type performance in some assays and improved perfor-mance in others. Several sites identified in previousCEL6A engineering efforts were included in the librariesand were identified again from the screen results.

ConclusionsBiomass-to-ethanol plants are being constructed andoperated today, but the foundation was laid by Mandels,

Reese and others in the 1950s and 1960s. In fact, Reesecredits Mandels with changing the focus of cellulaseresearch at the U.S. Army Natick Laboratory from ‘pre-vention of decomposition to promotion of decomposi-tion’ [55]. Reese and Mandels [24] demonstrated thatcellobiohydrolase activity was limiting cellulose hydroly-sis, but 30 years later, there are few reported successesin improving cellobiohydrolase-specific activity. Thereare many technical challenges to increasing cellulose-specific performance, including development of

Figure 4 Relative activity of CEL6A molecules. Purified CEL6A and purified CEL6A-His6 were compared in two cellulose hydrolysis activityassays. CEL6A and CEL6A-His6 exhibited similar activity in PASC hydrolysis (top graph). PASC activity was determined by measurement ofreducing sugars by PAHBAH following 1-hr incubation at 50°C in 1% (wt/wt) PASC. However, CEL6A-His6 activity was compromised, comparedto native CEL6A, in hydrolysis of dilute acid pretreated corn stover (bottom graph), PCS miniaturized assay.


of 13

Figure 5 Performance index. The performance index is the ratio of performance of the variant to the wild-type protein. PI >1 is improved overwild-type performance.

Figure 6 PASC assay controls. The sugar detected from the addition of the wild-type control CEL6A on a plate-by-plate basis plotted with aglobal curve fit (shown in the bottom right graph) to all of the data (second from the right at the bottom) collected during screening.


of 13

representative and predictive screens, expression of var-iants in an appropriate host, measurement of specificactivity which includes high-throughput specific proteindetermination, and the ability to query cellulase activityin a background of confounding activities. T. reesei wasdemonstrated to be an effective high-throughput screen-ing host for protein engineering. Specific activity screenswere developed and shown to detect improvements inCEL6A activity in a T. reesei background with real bio-mass substrates at intermediate solids loadings. Weimproved the thermal stability of the two H. jecorinacellobiohydrolases to the same level as CEL5A andCEL3A. CEL6A variants were identified with higheractivity than wild type and without loss of thermal stabi-lity. Since the methods used reflect process relevantconditions, they will help to quickly translate screeningsuccess to industrial success, without the potential pit-falls of changing substrate, solids loading, expressionhost, or protein background. While cost reductions willbe achieved through process optimization, improved cel-lulases and cellulase preparations will be needed tofurther reduce the cost of delivering cheap sugars to thebiofuels and biochemicals industries.

AbbreviationsCBH: Cellobiohydrolase; DDCE: DuPont Danisco Cellulosic Ethanol, LLC; DSC:Differential Scanning Calorimetry; EG: Endoglucanase; MTP: Microtiter plate;NREL: National Renewable Energy Laboratory; PASC: Phosphoric acid swollencellulose; PCS: Dilute acid pretreated corn stover; PI: Performance Index; SEL:Site Evaluation Library.

AcknowledgementsPortions of this work were funded by subcontract no. ZCO-0-30017-01 withthe National Renewable Energy Laboratory under prime contract no. DE-AC36-99G10337 with the U.S. Department of Energy or under award no: DE-FC36-08GO18078 awarded by the U.S. Department of Energy. Accordingly,the U.S. government may have certain rights. The authors would like toextend their appreciation to Dan Schell, Nancy Dowe-Farmer, and MikeHimmel at NREL; and to Roopa Ghirnikar and the Genencor project team inPalo Alto and Leiden.

Author details1Genencor Division, Danisco USA Inc., 925 Page Mill Rd. Palo Alto, CA 94304,USA. 2Genencor, a Danisco Division, Archimedesweg 30, 2333CN, Leiden,The Netherlands. 3Department of Molecular Biology, Swedish University ofAgricultural Sciences, POB 590, SE-751 24 Uppsala, Sweden.

Authors’ contributionsSEL conducted cellulase performance assays and mixture experiments, FGdeveloped assays and selected protein engineering sites, BRK developedassays and conducted mixture experiments, TK conducted His-tagcomparison and selected engineering sites, CM selected engineering sitesand provided technical leadership and coordination, RH coordinated andperformed high throughput screening, LW performed DSC analysis andselected engineering sites, JS solved protein structures and selected

Figure 7 Normal distribution of CEL6A controls. Activity values of the library recreated wild-type CEL6A, not used for curve fitting, weretransformed to determined PI values and further transformed with the natural log to obtain data that is normally distributed.


of 13

W ild type C e l6A ( )

-7

-6

-5

-4

-3

-2

-1

0

1

2

3

-5 -4 -3 -2 -1 0 1 2

ln (P I) P A S C

ln(P

I) PC

S

-7

-6

-5

-4

-3

-2

-1

0

1

2

3

-5 -4 -3 -2 -1 0 1 2

ln (P I) P A S C

ln(P

I) PC

S

Figure 8 Activity assay results. All of the wild-type and CEL6A variant activity data from the PASC and PCS assays was plotted. A correlation isobserved between the two activity assays. Improved variants in both assays are observed in the upper right quadrant. Wild type is shown inblue diamonds.

-3

-2

-1

0

1

2

3

4

5

6

7

8

-5 -4 -3 -2 -1 0 1 2

ln (P I) PASC

ln(P

I) he

at (

) a

nd E

tOH

(

)

-3

-2

-1

0

1

2

3

4

5

6

7

8

-5 -4 -3 -2 -1 0 1 2

ln (P I) PASC

ln(P

I) he

at (

) a

nd E

tOH

(

)

Figure 9 Stability assay results. All wild-type and CEL6A variant activity data from the PASC assay and both stability assays are plotted. Littlecorrelation is observed between activity and stability. Improved variants in both properties are observed in the upper right quadrant. Wild typeis shown in blue diamonds.


of 13

engineering sites, EAL developed assays and conducted mixtureexperiments. All authors read and approved the final manuscript.

Competing interestsThe authors declare that they have no competing interests.

Received: 15 April 2010 Accepted: 8 September 2010Published: 8 September 2010

References1. Reese ET, Siu RGH, Levinson HS: The biological degradation of soluble

cellulose derivatives and its relationship to the mechanism of cellulosehydrolysis. J Bacteriol 1950, 59:485-488.

2. Mandels M, Reese ET: Fungal cellulases and the microbial decompositionof cellulosic fabric. Devel Industrial Microbiol 1964, 5:5-20.

3. Caminal GJ, Sola LC: Kinetic modeling of the enzymic hydrolysis ofpretreated cellulose. Biotechnol Bioeng 1985, 27:1282-1290.

4. Eriksson T, Karlsson J, Tjerneld F: A model explaining declining rate inhydrolysis of lignocellulose substrates with cellobiohydrolase I (cel7A)and endoglucanase I (cel7B) of Trichoderma reesei. Appl BiochemBiotechnol 2002, 101:41-60.

5. Gonzalez GCG, De Mas C, Lopez-Santin J: A kinetic model for pretreatedwheat straw saccharification by cellulase. J Chem Technol Biotechnol 1989,44:275-288.

6. Eriksson T: Mechanism of surfactant effect in enzymatic hydrolysis oflignocellulose. Enzyme Microb Technol 2002, 31:353-364.

7. Gan Q, Allen SJ, Taylor G: Kinetic dynamics in heterogeneous enzymatichydrolysis of cellulose: an overview, an experimental study andmathematical modeling. Process Biochem 2003, 38:1003-1018.

8. Kadam KL, Rydholm EC, Knutsen JS, McMillan JD: Development andvalidation of a kinetic model for enzymatic saccharification oflignocellulosic biomass. Biotechnol Progress 2004, 20(3):698-705.

9. Rouvinen J, Bergfors T, Teeri T, Knowles JKC, Jones TA: Threee dimensionalstructure of cellobiohydrolase II from Trichoderma reesei. Science 1990,249:380-386.

10. Kraulis PJ, Clore GM, Nilges M, Jones TA, Pettersson G, Knowles J,Gronenborn AM: Determination of the three-dimensional solutionstructure of the carboxyl-terminal domain of cellobiohydrolase I fromTrichoderma reesei: a study using NMR and hybrid distance geometry-dynamical simulated annealing. Biochemistry 1989, 28:7241-7257.

11. Divne C, Stahlberg J, Reinikainen T, Rouhonen L, Pettersson G, Knowles JKC,Teeri TT, Jones TA: The three-dimensional crystal structure of thecatalytic core of cellobiohydrolase I from Trichoderma reesei. Science1994, 265:524-528.

12. Cantarel BL, Coutinho PM, Rancurel C, Bernard T, Lombard V, Henrissat B:The Carbohydrate-Active Enzymes Database (CAZy): an expert resourcefor glycogenomics. Nucleic Acids Res 2009, 37:D233-D238.

13. Meinke A, Damude HG, Tomme P, Kwan E, Kilburn DG, Miller RC Jr, RWarren AJ, Gilkes NR: Enhancement of the endo-b-1,4-glucanase activityof an exocellobiohydrolase by deletion of a surface loop. J Biol Chem1995, 270:4383-4386.

14. Bu L, Beckham GT, Crowley MF, Chang CH, Matthews JF, Bomble YJ,Adney WS, Himmel ME, Nimlos MR: The energy landscape for theinteraction of the family 1 carbohydrate-binding module and thecellulose surface is altered by hydrolyzed glycosidic bonds. J Phys ChemB 2009, 113:10994-11002.

15. Zhong L, Matthews JF, Hansen PI, Crowley MF, Cleary JM, Walker RC,Nimlos MR, Brooks CL, Adney WS, Himmel ME, Brady JW: Computationalsimulations of the Trichoderma reesei cellobiohydrolase I acting onmicrocrystalline cellulose Ib: the enzyme-substrate complex. CarbohydrRes 2009, 344:1984-1992.

16. Igarashi K, Koivula A, Wada M, Kimura S, Penttilä M, Samejima M: Highspeed atomic force microscopy visualizes processive movement ofTrichoderma reesei cellobiohydrolase I on crystalline cellulose. J Biol Chem2009, 284:36186-36190.

17. Imai T, Boisset C, Samejima M, Igarashi K, Sugiyama J: Unidirectionalprocessive action of cellobiohydrolase Cel7A on Valonia cellulosemicrocrystals. FEBS Lett 1998, 432(3):113-116.

18. Varrot A, Frandsen TP, von Ossowski I, Boyer V, Cottaz S, Driguez H,Schülein M, Davies GJ: Structural basis for ligand binding and processivity

in cellobiohydrolase Cel6A from Humicola insolens. Structure 2003,11:855-864.

19. Boisset C, Fraschini C, Schülein M, Henrissat B, Chanzy H: Imaging theenzymatic digestion of bacterial cellulose ribbons reveals the endocharacter of the cellobiohydrolase Cel6A from Humicola insolens and itsmode of synergy with cellobiohydrolase Cel7A. Appl Environ Microbiol2000, 66:1444-1452.

20. Moran-Mirabal JM, Santhanam N, Corgie SC, Craighead HG, Walker LP:Immobilization of cellulose fibrils on solid substrates for cellulasebinding studies through quantitative fluorescence microscopy. BiotechnolBioeng 2008, 101:1129-1141.

21. Igarashi K, Uchihashi T, Koivula A, Wada M, Penttilä M, Ando T, Samejima M:Single molecular observations of processive glycosidases on crystallinesubstrates. 14th Annual SBNet Meeting June 11-14, 2010 abstract, Tällberg,Sweden .

22. Baker JO, Ehrman CI, Adney WS, Thomas SR, Himmel ME: Hydrolysis ofcellulose using ternary mixtures of purified cellulases. Appl BiochemBiotechnol 1998, 70-72:395-403.

23. Rosgaard L, Pedersen S, Langston J, Akerhielm D, Cherry JR, Meyer AS:Evaluation of minimal Trichoderma reesei cellulase mixtures ondifferently pretreated barley straw substrates. Biotechnol Prog 2007,23:1270-1276.

24. Reese ET, Mandels M: Stability of the cellulase of Trichoderma reeseiunder use conditions. Biotechnol Bioeng 1980, 22(2):323-335.

25. Suominen PL, Mäntylä AL, Karhunen T, Hakola S, Nevalainen H: Highfrequency one-step gene replacement in Trichoderma reesei. II. Effects ofdeletions of individual cellulase genes. Mol Gen Genet 1993, 241(5-6):523-530.

26. Converse AO, Matsuno R, Tanaka M, Taniguchi M: A model of enzymeadsorption and hydrolysis of microcrystalline cellulose with slowdeactivation of the adsorbed enzyme. Biotechnol Bioeng 1988, 32:38-45.

27. Gusakov AV, Sinitsyn AP: A theoretical analysis of cellulase productinhibition: effect of cellulase binding constant, enzyme/substrate ration,and b-glucosidase activity on the inhibition pattern. Biotechnol Bioeng1992, 40:663-671.

28. Holtzapple M, Cognata M, Shu Y, Hendrickson C: Inhibition of Trichodermareesei cellulase by sugars and solvents. Biotechnol Bioeng 1990, 36:275-287.

29. Jeoh T, Ishizawa CI, Davis MF, Himmel ME, Adney WS, Johnson DK:Cellulase digestibility of pretreated biomass is limited by celluloseaccessibility. Biotechnol Bioeng 2007, 98(1):112-122.

30. Mandels M, Weber J, Parizek R: Enhanced cellulase production by amutant of Trichoderma viride. Appl Environ Microbiol 1971, 21(1):152-154.

31. Schimenti J, Garrett T, Montenecourt BS, Eveleigh DE: Selection ofhypercellulolytic mutants of Trichoderma reesei based on resistance tonystatin. Mycologia 1983, 75:876-880.

32. Sheir-Neiss G, Montenecourt BS: Characterization of the secretedcellulases of Trichoderma reesei wild type and mutants during controlledfermentations. Appl Microbiol Biotechnol 1984, 20:46-53.

33. Jalak J, Väljamäe P: Mechanism of initial rapid rate retardation incellobiohydrolase catalyzed cellulose hydrolysis. Biotechnol Bioeng 2010,106(6):871-883.

34. Schülein M: Protein engineering of cellulases. Biochim Biophys Acta 2000,1543(2):239-252.

35. Koivula AT, Kinnari T, Harjunpää V, Ruohonen L, Teleman A, Drakenberg T,Rouvinen J, Jones TA, Teeri TT: Tryptophan 272: an essential determinantof crystalline cellulose degradation by Trichoderma reeseicellobiohydrolase Cel6A. FEBS Lett 1998, 429(3):341-346.

36. Wohlfahrt G, Pellikka T, Boer H, Teeri TT, Koivula AT: Probing pH-dependentfunctional elements in proteins: modification of carboxylic acid pairs inTrichoderma reesei cellobiohydrolase Cel6A. Biochemistry 2003,42:10095-10103.

37. Koivula A, Reinikainen T, Ruohonen L, Valkeajärvi A, Claeyssens M,Teleman O, Kleywegt GJ, Szardenings M, Rouvinen J, Jones TA, Teeri TT:The active site of Trichoderma reesei cellobiohydrolase II: the role oftyrosine 169. Protein Eng 1996, 9:691-699.

38. Koivula A, Ruohonen L, Wohlfahrt G, Reinikainen T, Teeri TT, Piens K,Claeyssens M, Weber M, Vasella A, Becker D, Sinnott ML, Zou J-Y,Kleywegt GJ, Szardenings M, Ståhlberg J, Alwyn Jones T: The active site ofcellobiohydrolase Cel6A from Trichoderma reesei: the roles of asparticacids D221 and D175. J Am Chem Soc 2002, 124:10015-10024.


of 13

http://www.ncbi.nlm.nih.gov/pubmed/15436422?dopt=Abstract







































































39. Zou J-Y, Kleywegt GJ, Ståhlberg J, Driguez H, Nerinckx W, Claeyssens M,Koivula A, Teeri TT, Alwyn Jones T: Crystallographic evidence for substratering distortion and protein conformational changes during catalysis incellobiohydrolase Cel6A from Trichoderma reesei. Structure 1999,7(9):1035-1044.

40. Zhang S, Irwin DC, Wilson DB: Site-directed mutation of noncatalyticresidues of Thermobifida fusca exocellulase Cel6B. Eur J Biochem 2000,267:3101-3115.

41. Ai Y-C, Zhang S, Wilson DB: Positional expression effects of cysteinemutations in the Thermobifida fusca cellulose Cel6A and Cel6B catalyticdomains. Enz Microbial Technol 2003, 32:331-336.

42. Wilson DB: Cellulases and biofuels. Curr Opin Biotechnol 2009, 20:295-299.43. Heinzelman P, Snow CD, Wu I, Nguyen C, Villalobos A, Govindarajan S,

Minshull J, Arnold FH: A family of thermostable fungal cellulases createdby structure-guided recombination. Proc Natl Acad Sci USA 2009,106:5610-5615.

44. Heinzelman P, Snow CD, Smith MA, Yu X, Kannan A, Boulware K,Villalobos A, Govindarajan S, Minshull J, Arnold FH: SCHEMA recombinationof a fungal cellulase uncovers a single mutation that contributesmarkedly to stability. J Biol Chem 2009, 284:26229-26233.

45. Mandels M, Andreotti R, Roche C: Measurement of saccharifying cellulase.Biotech Bioeng Symp 1976, 6:21-33.

46. Doner LW, Irwin PL: Assay of reducing end-groups in oligosaccharidehomologues with 2,2’-bicinchoninate. Analyt Biochem 1992, 202:50-53.

47. Miller GL: Use of dinitrosalicylic acid reagent for determination ofreducing sugar. Anal Chem 1959, 31:426-428.

48. Ghose TK: Measurement of cellulase activities. Pure Appl Chem 1987,59(2):257-268.

49. Lever MA: New reaction for colorimetric determination of carbohydrates.Anal Biochem 1972, 47(1):273-279.

50. Kuwajima K: Circular dichroism. Methods Mol Biol 1995, 40:115-35.51. Woody RW: Circular dichroism. Methods Enzymol 1995, 246:34-71.52. Kelly SM, Price NC: The application of circular dichroism to studies of

protein folding and unfolding. Biochim Biophys Acta 1997,1338(2):161-185.

53. Wood T: Biomass part A: cellulose and hemicellulose. In Methods inEnzymology. Edited by: Wood W, Kellog S. San Diego, Academic Press;1988:19-25.

54. Schell DJ, Farmer J, Newman M, McMillan JD: Dilute-sulfuric acidpretreatment of corn stover in pilot-scale reactor: investigation of yields,kinetics, and enzymatic digestibilities of solids. Appl Biochem Biotechnol2003, 105(1-3):69-85.

55. Reese ET: History of the cellulase program at the U.S. Army NatickDevelopment Center. Biotechnol Bioeng Symp 1976, 6:9-20.

doi:10.1186/1754-6834-3-20Cite this article as: Lantz et al.: Hypocrea jecorina CEL6A proteinengineering. Biotechnology for Biofuels 2010 3:20.

Submit your next manuscript to BioMed Centraland take full advantage of:

• Convenient online submission

• Thorough peer review

• No space constraints or color figure charges

• Immediate publication on acceptance

• Inclusion in PubMed, CAS, Scopus and Google Scholar

• Research which is freely available for redistribution

Submit your manuscript at www.biomedcentral.com/submit


of 13

























research open access hypocrea jecorina cel6a protein ... · research open access hypocrea jecorina...

Documents